Lung cancer is among the best known examples of a disease that reflects the interaction of genetic and environmental factors. While carcinogenic compounds in tobacco smoke constitute the major risk factor for lung cancer, only about 14% of long-time smokers will develop cancer. Family studies indicate the role that genetic susceptibility plays in determining lung cancer risk. The goal of this research proposal is to develop and apply methods to identify genetic susceptibility factors for lung cancer that are modulated by the effects of tobacco smoke. Toward this goal we are taking advantage of two extensive sources of data that reflect the efforts of teams of scientists who have been devoted to understanding the causes of lung cancer. The first source of data is from the Genetic Epidemiology of Lung Cancer Consortium, which has been collecting extended families with three or more relatives affected with lung cancer since 2000. Analytical results show evidence for a genetic susceptibility factor on chromosome 6q that strongly increases lung cancer risk, but that has a much more profound effect on risk among tobacco smokers. The other major source of data is an extremely large, case-control study developed by Dr. Margaret Spitz at the U.T. M.D. Anderson Cancer Center, starting in 1993. Currently, candidate gene studies to identify genetic risk factors of lung cancer susceptibility have been completed for 51 polymorphisms. In addition, data from a newly launched initiative to perform a genome-wide association analysis of 1200 ever-smoking Caucasian lung cancer cases and 1200 matched controls will become available within the next 3 months. A major goal of this proposal is to develop simulation-based approaches to evaluate the efficacy of many competing analytical approaches for characterizing the role of gene-environment interactions in disease causation. Our research involves a strong analytical team with extensive experience in gene-environment modeling, application of machine learning tools, and the study of family data. In addition, we have unique skills and methods for simulating data, which will be further refined through the proposed research funding. The simulation approaches and analytical discoveries from our research will be widely distributed. (End of Abstract)